In the process of controlling a moving object, initial direction of the moving object relative to the prevailing direction should first be fixed. For many tasks relating to vehicle transportation management, it is sufficient to determine just a direction in which a vehicle starts to move—“forward” or “backward” relative to the prevailing movement direction, which is mostly used for the given vehicle considering its design and/or its operational specifics.
For example, a four-wheel vehicle, which turns by rotating one wheel pair, has a prevailing moving direction (the most likely used) which passes through the center of the second wheel pair towards the center of the first (steering) wheel pair.
If an object, such as a four-wheel tractor, moves forward and needs to turn right, the steering system turns the wheels to a positive angle (to the right); if the tractor moves backward (uses a reverse gear) and needs to turn to the right relative to the current moving direction, then the wheels should be turned to a negative angle (to the left), and some auxiliary elements (swiveling mirrors, reversing camera, parktronic, audible warning etc) should be used.
A method of vehicle moving direction estimation using transmission control information according to U.S. Pat. No. 7,451,029 is implemented by mechanical connection of a device detecting the vehicle's movement to a transmission controller or vehicle wheels.
The drawback of this method is the necessity of the mechanical connection between a special detector with engine transmission or vehicle's wheels, as well as a possible mismatch in the direction of wheel rotation and real vehicle movement, for example, due to wheel slip or frictional sliding.
An orientation measurement apparatus and method according to U.S. Pat. No. 4,644,358 is implemented by the use of a set of rotating antennas receiving Global Navigation Satellite System (GNSS) signals.
The drawback of this method is the necessity of using a complex rotating antenna and a separate computing device, as well as high sensitivity to measurement errors and interference.
As an alternative, an accelerometer (e.g., a one-axis accelerometer, or a three-axis accelerometer, or a MEMS-type accelerometer) can be used, although the accuracy of most accelerometers is only good for rapidly accelerating vehicles, and often relatively poor for slowly moving vehicles. Most commercial accelerometers have relatively high errors in accuracy. Highly accurate accelerometers are often expensive, and not always available for non-military applications.
As another alternative, an inertial measurement unit (e.g., 6 degrees of freedom IMU, consisting of three accelerometers and three gyroscopes) can be used, but usually robustness/reliability of direction detection for slowly moving vehicles is not enough.
Sensitive axis of accelerometer should be placed along longitudinal axis of machine and pointed, e.g., in a forward direction. In the ideal case, when the vehicle starts to move forward, the accelerometer will output positive value. When backward—the accelerometer will output a negative value. The sign of the measured acceleration can be used for direction detection. However, a typical vehicle moving, e.g., forward usually has unstable acceleration, which has both positive and negative pulses that do not allow to properly detect direction. On the other hand, an accelerometer installed on the vehicle with a suspension system may be affected by pitching/rocking caused by a moment of inertia, which leads again to unstable fluctuating acceleration.
Moreover, the accelerometer itself has two types of errors: one changes fast and is called “noise”, the other changes slowly and is called “bias”. These issues do not allow to properly recognize actual movement, and usually an approach of combining accelerometer with GNSS receiver measurement is used.
Integrating techniques are usually based on Kalman filtering, which is well known approach in the art. Kalman filter (KF) allows to process different type of measurements simultaneously, taking into account their inherent errors and estimate parameter of movement such as position, velocity and acceleration. To properly work, the filter should align GNSS coordinate system and inertial coordinate system, i.e., the system used by the accelerometer. As long as the direction is not resolved, there can be two hypotheses: forward and backward movement. The best parameter to check the hypothesis is velocity. FIG. 11 shows velocity measured by GNSS and estimated by KF for a fast moving object in case of true hypothesis about movement direction. Maximum velocity about 7 m/s or 25 km/h. One can see in such case difference between measured and estimated values has behavior of noise which do not exceed 0.5 m/s threshold. FIG. 12 shows case of wrong hypothesis for the same movement scenario. One can see here that the difference has a large spiked behavior which exceed 0.5 m/s. Comparison of difference with threshold can be used as a criteria for direction detection. FIG. 13 shows a case of a slowly moving object with velocity 0.6 m/s or 2 km/h. There in case of true hypothesis difference is within threshold of 0.5 m/s. However in case of wrong hypotheses (see FIG. 14), the difference is again within thresholds. So, the direction detection based on the accelerometer does not work for the case of slowly movement object.
Also, because of its construction, the accelerometer does not detect pure acceleration but also Earth's gravity. Therefore, moving on an uneven surface or just changing of pitch of the moving object will add a projection of gravity vector to acceleration measurement and will need separate pitch or gravity vector measurement to calculate acceleration from accelerometer measurements.
Thus, the objective is finding approach that allows to properly detect direction independent of object structure and dynamics of object movement.